A reprint from
American Scientist
the magazine of Sigma Xi, The Scientific Research Society
This reprint is provided for personal and noncommercial use. For any other use, please send a request to Permissions,
American Scientist, P.O. Box 13975, Research Triangle Park, NC, 27709, U.S.A., or by electronic mail to [email protected].
©Sigma Xi, The Scientific Research Society and other rightsholders
ENGINEERING
ST. FRANCIS DAM
Henry Petroski
os Angeles could not have grown into the
metropolis that it is today without the expansion of its water supply. In 1900 the city’s
population was about 100,000 and growing
rapidly, to reach 175,000 within five years. Since
the Los Angeles River watershed was capable of
supporting only about 200,000 people, the city
had the choice of limiting growth or finding new
sources of water. A drought in 1904 raised the issue to crisis proportions.
Los Angeles’s need to import water had been
foreseen a decade earlier by Fred Eaton, who as
city engineer had identified the Owens Valley,
north of the city, as a likely candidate. In the
meantime, the U.S. Reclamation Service had begun looking into the feasibility of an irrigation
scheme for the farmers of the Owens Valley, and
Los Angeles had to act fast if it was to obtain the
water rights. Eaton took William Mulholland,
manager of the newly formed Los Angeles Bureau of Water Works and Supply, to the valley to
investigate the possibility of constructing a gravity-flow aqueduct from there to the city nearly
250 miles south. The distance was unprecedented. The longest Roman aqueducts were less than
60 miles long, and New York’s Croton Aqueduct
was even shorter. However, Owens Lake was
more than 3,000 feet higher than the city, providing a much greater average gradient than existed
in the successful Croton Aqueduct. Thus the engineering problems, which would involve inverted siphons and pressure tunnels to get the
water over and through the mountains in the
way, seemed solvable.
L
Mulholland Drive
William Mulholland was an engineer of the old
school, which essentially means that he had learned
by doing. He was born in Ireland in 1855, went to
sea at 15, landed in New York City four years later,
worked at a variety of jobs in the East and Midwest, and sailed via the Isthmus of Panama to San
Francisco. He settled in the Los Angeles area at the
age of 22, working as a “water ditch tender” with
the Los Angeles City Water Company, a small private provider. According to one account:
Henry Petroski is A. S. Vesic Professor of Civil Engineering and a
professor of history at Duke University. Address: Box 90287,
Durham, NC 27708-0287.
114
American Scientist, Volume 91
Mulholland later recalled that he became interested in things technical when serving as
a helper on a drill rig digging water wells
that pierced a buried tree trunk at a depth of
600 feet. He went to the library to investigate the manner by which a tree could become buried at such great depth, and read
University of California Professor John
LeConte’s Introduction to Physical Geology.
Mulholland liked the subject matter so
much that he later recalled, “Right there I
decided to become an engineer”.…
Mulholland eventually became general manager
and chief engineer of the Los Angeles Water Company. He proved himself to be so knowledgeable of
the poorly documented infrastructure and workings of the water distribution system that, when
the company was acquired by the city in 1902, the
self-taught engineer was retained as its manager. It
was in this capacity that he accompanied Fred
Eaton to the Owens Valley and secured $1.5 million
from the Los Angeles Board of Water Commissioners for engineering studies of the situation.
The scale of the project and Mulholland’s “lack
of substantive experience in constructing such facilities” were used by “other engineers, newspaper editors and electric power interests” to discredit the scheme. In response to the criticism, the
City Commissioners appointed an Aqueduct Advisory Board, comprising three distinguished consulting engineers, to “make an independent evaluation of the proposed aqueduct design.” One of
the consultants was John R. Freeman, who had
been among the principal designers of the New
Croton Aqueduct and who had served on the advisory board to review the design of the Panama
Canal. When the external board found Mulholland’s aqueduct design “admirable in conception
and outline,” criticism was quelled. The $23 million bond issue passed overwhelmingly in 1907.
The construction of the 233-mile aqueduct and
its initial filling were not without incident. Upon
first carrying water, one of the major siphons in
the aqueduct began leaking and was lifted up by
the resulting hydraulic forces. The seepage from
the riveted steel conduit also triggered a landslide in which the pipe became entangled. Such
setbacks were forgotten by most of the 30,000
people who gathered on November 5, 1913, to
© 2004 Sigma Xi, The Scientific Research Society. Reproduction
with permission only. Contact [email protected].
Figure 1. St. Francis dam impounded 38,000 acre-feet of water prior to its failure on March 12, 1928. (Photograph by the Los Angeles Department of Water and Power.)
watch the opening of the world’s longest aqueduct, which was capable of transporting 258 million gallons per day to Los Angeles.
While the aqueduct was being planned, speculators bought up large parcels of land in the San Fernando Valley, located north of Los Angeles. Water
from the aqueduct would make the semi-arid region arable, they anticipated, but when it became
clear that no such water would ever be made available outside of Los Angeles, the San Fernando
landowners argued for annexation. By 1924, their
successful campaign had quadrupled the area of
the city. This growth, combined with a three-year
drought, severely taxed the water supply. Under
the worst conditions, ranchers in the San Fernando
Valley were intercepting virtually all of the aqueduct’s base flow. The city of Los Angeles sought to
acquire more water rights in the Owens Valley, but
angry residents, still bitter from the original conflict
over the claims of the rural valley versus a growing
city, balked, and some turned to violence reminiscent of the Old West. Among the retaliatory acts
was the dynamiting of the aqueduct, which subsequently had to be protected by armed guards.
Storage to Stretch Supply
In the meantime, in recognition of the fact that the
aqueduct could not supply enough water for both
urban Los Angeles and the rural San Fernando
Valley without enormous storage capacity, additional reservoirs had been planned and designed
and were under construction. In fact, between 1920
and 1926, a total of eight new reservoirs were built
by the Los Angeles Bureau of Waterworks and
Supply, during which time Mulholland made it
known that it was his goal to have enough reservoir capacity to hold in reserve an entire year’s
worth of water for the city. Among the additional
reservoirs Mulholland planned was one that
would account for about half the total water required. This dam, to be located in San Francisquito
Canyon, was to be called the St. Francis.
The St. Francis Dam site was chosen after inflated land values made another location too ex-
pensive for Mulholland’s tastes. In fact, he had
imagined a dam in San Francisquito Canyon during the construction of the aqueduct. Mulholland
saw then that a relatively small dam built where
the canyon narrowed would hold back an enormous amount of water. He also recognized early
on that the geology of the location called for special caution, but these conditions did not keep
him from designing a dam for the site. He assumed that the buttressing effect of the dam
would mitigate any slippage at the canyon walls.
Until 1923, all the dams whose designs Mulholland had overseen were earthworks—large embankments whose fine-grained silt and clay cores
were more or less impermeable to water. The first
concrete dam built for Los Angeles was the 200foot-high Weid Canyon Dam, which was designed to impound the Hollywood Reservoir. It
has been speculated that Mulholland decided to
adopt a concrete-dam design over the clay-core
type with which he was so familiar because of the
limited supply of clayey materials in the sides of
Weid Canyon. A year before the unique concrete
dam was completed, in 1925, it was christened
Mulholland Dam, a testament to the stature to
which the chief engineer had risen in Los Angeles.
St. Francis Dam was similarly designed to be
made of concrete, because there was no suitable
clay or silt available at the San Francisquito
Canyon site. The new dam would also be a
stepped concrete gravity arch structure: Its downstream face was constructed like a wide set of
steps, its material was mass (unreinforced) concrete, and the structural principle by which it held
back the water was through its sheer weight pressing down on the ground, aided by an arched plan
that took advantage of the water pressure behind
it to compress or wedge the dam between the
sides of the canyon, which served as abutments.
The original design of the St. Francis called for a
dam reaching 175 feet above the San Francisquito
Creek bed, which would have given it a capacity
of 30,000 acre feet of water—that is, enough water
to flood 30,000 acres to a depth of one foot, enough
© 2004 Sigma Xi, The Scientific Research Society. Reproduction
with permission only. Contact [email protected].
2003
March–April
115
to supply Los Angeles for a year. But because of
increased water use by Los Angeles, before the
first concrete was poured, the dam’s capacity was
increased to 32,000 acre feet by raising its height
and adding a wing dike that extended from the
west abutment. Almost a year after the beginning
of the placement of concrete, apparently in response to further growth in water usage, the reservoir capacity was once again increased by raising
and extending the wing dike and by adding another 10 feet to the dam’s height, to increase its
capacity to more than 38,000 acre feet—more than
25 percent greater than the original design. The
changes in its height had been made without a
proportionate widening of the dam’s base, but
Mulholland believed that the design still had a
“factor of safety of three or four.” A gravity dam
derives its ability to hold back water without tipping over from the width of its base, however, so
the factor of safety of the dam was definitely reduced by the design changes.
Less Than Conservative Design
Construction of St. Francis Dam lagged that of
Mulholland Dam by about a year, and the successful advance of that structure must have provided
plenty of confidence in the safety and robustness of
the basic design, in spite of some less than conservative design features. St. Francis Dam contained
130,000 cubic yards of concrete but no reinforcing
steel. The main structure also lacked contraction
joints, which allow concrete to crack in a controlled
manner as it cools. (The grooves in a concrete sidewalk cause it to crack at the base of these reduced
sections, thus keeping the predictable cracks more
or less straight and hidden.) No doubt the arched
nature of the dam was expected to close as much
as possible any cracks that did develop. St. Francis
Dam was also constructed without drainage galleries, tunnels that run through a structure to allow
inspection for cracks and sources of leakage, and to
provide a means for monitoring the amount of water flowing through the dam. Finally, the structure
was built without cut-off walls (concrete-filled
trenches designed to reduce water seepage under
the dam) or a grout curtain (a further seepage-prevention measure taken by forcing grout under
pressure into holes drilled in the rock under a cutoff wall). These measures reduce the possibility of
water infiltrating under a dam and exerting upward hydrostatic pressure, thus making the structure somewhat buoyant. Such buoyancy provides
an uplift force that reduces the effect of the weight
of the dam in keeping it in place. In extreme cases,
uplift forces can cause a dam to tend to tilt forward
or slide downstream. In short, many of the design
features of St. Francis Dam, which were in accordance with standard engineering practice of the
time, in retrospect contributed to making it less watertight, less inspectable and less stable than should
have been considered wise.
By the time he was building St. Francis, Mulholland’s record of successful dams appears to have
116
American Scientist, Volume 91
given him a confident attitude toward the ability of
the gigantic structures to hold back the force of water, but at the same time, he had resigned himself
to the fact that some water leaked through. Although his experience was with earthen dams, he
evidently felt comfortable transferring his confidence to concrete dams, which, after all, were
made of a stronger and less permeable material.
St. Francis Dam was completed in May 1926, but
months before that date water from the Owens
aqueduct was diverted into the reservoir. At first,
enough water was allowed to pass through outlets
in the dam to maintain the flow in San Francisquito
Creek. Shortly after St. Francis Dam was completed, however, Los Angeles requested the appropriation also of “flood and surplus waters,” and
blocked the flow into San Francisquito Creek. Mulholland is said to have believed that the Santa Clarita Valley ranchers downstream could continue to
draw water from their wells, not appreciating that
the replenishment of the groundwater depended
on the creek flow. An agreed-upon test release of
water from the dam caused the resulting stream to
dry up within a few miles, indicating that the water
indeed was going into the ground. This incident
was one of Mulholland’s few public embarrassments over water issues, but it also demonstrated
that his assertions about water flow through the
ground were not always fully informed.
A year after St. Francis Dam was completed,
the level of its reservoir reached within three feet
of the crest of the spillway, which was designed to
keep water from overflowing the top of the dam.
The water did not reach the spillway, however, for
the spring runoff ceased and the level of the reservoir began to drop. The cracks that had developed
in the dam during its filling were described by
Mulholland as “transverse contraction cracks”
and did not appear to alarm him. The downstream crevices were “infilled with hemp and
sealed with wedges of oakum” and “backfilled
with cement grout to seal off active seepage.”
The next year’s spring runoff caused the reservoir to fill again, this time to maximum capacity. In
the meantime, new leaks developed in the dam,
some manifesting themselves in springs in the
foundation and others in the old cracks—through
which the discharge was increased over the previous year. Still other leaks developed on either abutment of the dam and in the wing dike. Mulholland
ordered a concrete pipe installed to drain water
from this last leak toward the abutment of the dam.
Disaster
March 12, 1928—75 years ago—was a windy day,
and water from the reservoir was blown in waves
against the dam and over its spillways. This water
naturally washed over the stepped downstream
face of the dam, making it difficult to tell if new
leaks were developing or old ones were growing.
Full reservoirs throughout the system and winter
runoff combined to present an abundance of water, which was allowed to flow into San Francis-
© 2004 Sigma Xi, The Scientific Research Society. Reproduction
with permission only. Contact [email protected].
quito Canyon for the first time in almost two
years. Earlier that day, the St. Francis damkeeper
had called Mulholland to bring to his attention a
new leak, of “dirty” water, at the west abutment.
Such water could indicate that foundation material was being washed out from under the dam,
which could lead to it being undermined—certainly a dangerous condition. Mulholland, who
claimed to have made it a practice to visit all 19
dams under his supervision at least once every
two weeks, immediately drove with his assistant
to the St. Francis, where they spent two hours inspecting the dam. The “dirty” water seemed clear
to them, and the dam was declared safe.
A little before midnight that same day, the dam
gave way and the contents of the reservoir inundated San Francisquito Canyon. The water rushed
down the canyon, destroying everything in its
path. Some large sections of the concrete dam,
weighing thousands of tons, were washed as much
as a mile downstream, leaving only the tall center
section of the structure remaining in place, with
some other large blocks scattered nearby, mostly
between the center section and what was the east
abutment. A powerhouse located about a mile and
a half downstream was washed away, as were a
construction camp and houses in little towns and
villages in the path of the water. Hundreds of people were killed, most no doubt unsuspecting as
they slept. The official death toll was in excess of
430, but the actual total is debated to this day.
According to Engineering News-Record, the magazine of the construction industry that had long
ago established its reputation for accurate and incisive reporting on the failure of structures, it was
“the first time in history a high dam of massive
masonry” had failed. The disaster was compared
to the Johnstown Flood of 1889, which had claimed
more than 2,000 lives, a disaster that was once considered “the worst in history resulting from failure of manmade structures.” However, the trade
magazine declared, “the washing out of an old neglected earth dam was not an engineering
tragedy” so much as a case of carefree modifications and poor maintenance by the hunting and
fishing club that had patched an abandoned Pennsylvania state canal system reservoir to make a
recreational lake. The failure of the St. Francis Dam
was indeed different, for here the latest engineering materials, design philosophies, construction
techniques and operational procedures were overseen by an engineer with an impeccable record of
success. That is not to say that Mulholland’s work
was without critics, but in the business of holding
back vast quantities of water, he had been able to
answer their fears. At least in the minds of those
who were in the position to give the go-ahead for
such great projects, a concrete dam built by Mulholland would certainly be stronger than the water
that pushed against it. But, as an editorial in Engineering News-Record put it, “Men have always been
in awe of these vast forces, and often has bitter
protest been made against the erection of a dam
Figure 2. William Mulholland (left) views the remains of St. Francis
dam. (Photograph courtesy of the Wilkman Productions Collection.)
above populous communities. In every instance
engineering science answered the protest and gave
assurance that the waters would be safely controlled. The destruction of the St. Francis Dam challenges that assurance.”
Devilish Details
A great failure is the perfect counterexample to a
hubristic hypothesis. William Mulholland and
his staff had evidently so gained confidence in
their mastery of the great hydraulic forces pent
up behind the successful dams they had built
that they began to build them with less and less
attention to detail, especially the all-important local detail of the geology underlying the site. Or, if
they did pay close attention to it, they missed
some key elements of its character.
More than a dozen official boards and commissions were appointed by various California offices
and interested parties, ranging from the state governor to the Los Angeles County district attorney, to
investigate the St. Francis Dam disaster. Within a
month or so of the incident, studies were made,
witnesses were interviewed, hearings were held,
and a half dozen reports were filed. All six of these
identified the foundations of the dam to have been
inadequate, but there was no unanimity over the
exact triggering mechanism for the catastrophe.
The reports dismissed early speculations about
an earthquake or explosion causing any initial
breach. They also confirmed that the concrete was
of sufficient strength, and that the design of the superstructure was in accordance with commonly accepted engineering practice. (The failure appears
not to have had any significant effect on the structural design of concrete dams at the Bureau of
© 2004 Sigma Xi, The Scientific Research Society. Reproduction
with permission only. Contact [email protected].
2003
March–April
117
Reclamation, which had succeeded the Service in
1923.) However, the reports condemned the foundations underlying the St. Francis Dam, a subject
that in general was incompletely understood at the
time. Among the things that were pointed out in
the various reports were that the conglomerate material under the upper right abutment structure
was held together with clay that softened when
wet and furthermore contained numerous fissures
filled with gypsum, which dissolves in water. Any
water flowing through this material would in time
carry out gypsum, which would not be easily seen,
for the water would appear to be clear. When
enough gypsum had dissolved and the foundations were sufficiently softened, the dam would
settle unevenly, eventually crack and finally have
no ability to hold back the water behind it, which
would rush freely downstream.
The reports were in general agreement that it
was the west side of the dam, where the underlying rock was as described, that gave way first. The
east side of the dam had abutted the steep slope of
the canyon that consisted of parallel layers of
schist. A landslide that was evident at the site of
the failed dam was generally assumed to have occurred after the dam was breached. Geological engineer J. David Rogers, however, maintains that it
was a landslide of the east canyon wall that triggered the failure, with the rush of water scouring
the dam’s foundation and causing the dam to lean.
This in turn opened up preexisting cracks all along
the arched structure. According to Rogers’ recent
studies, the landslide dumped as much as a million cubic yards of weathered mica schist, which
“created an outpouring flood wave supercharged
with sediment.” If the density of such sediment
was about five times that of water, Rogers believes
that it could have effectively made large sections of
the dam sufficiently buoyant to be pushed and
tumbled down the canyon, where they were found
in the wake of the ensuing flood. This hypothesis,
like others, remains unprovable in any rigorous
sense, of course, but it highlights the complexity of
anticipating the forces on a concrete dam structure.
Regardless of the exact mechanism by which
the St. Francis dam cracked and gave way,
William Mulholland took full responsibility for
the disaster. On the witness stand, he admitted
that he could not explain the failure. According
to Engineering News-Record, “He had based his
opinion as to its safety on previous experience in
building nineteen dams. His only suggestions as
to the possible cause was that ‘we must have
overlooked something.’” Overlooking something
is, of course, always a danger in the design of
large engineering systems, and it is precisely why
the opinion of independent experts is sought
during the design stage. In addition to the technical error of siting the St. Francis Dam on poor
foundations, its collapse was blamed on the “human factor,” which manifested itself in the fact
that “engineering work in the Bureau of Water
Works and Supply always had been dominated
118
American Scientist, Volume 91
by one man, the chief engineer, who took upon
himself, in this case at least, entire responsibility,
sought no independent opinions and adopted
technical policies based on his unconfirmed judgment alone.” Since “higher officials had absolute
confidence in Mr. Mulholland,” outside opinion
was not sought and “there was no intervention
from above.” The plans for the dam were not
challenged because “Mr. Mulholland was personally overseeing the work.”
No engineer should have such hubris as to
think that past successes are sufficient to guarantee the success of the next project. Each new
project rests on a new foundation, whose hidden
faults may or may not be within prior experience.
When all dams and the foundations upon which
they rest begin to look alike to an engineer like
William Mulholland, he himself should question
his own expertise. As Engineering News-Record
put it just two months after the disaster, “Had
the plan of construction used for the St. Francis
Dam been brought forward by some comparatively inexperienced engineer, or had the work
been done by contract or under any other condition that would naturally have brought independent engineering opinion into the case, it is highly probable that some modified plan would have
been substituted and the disaster avoided.” Experience is not always the best teacher.
Acknowledgment
I am grateful to Jon Wilkman for sharing with me his
proposal to make a documentary film on the St.
Francis Dam disaster, for providing background
material and photographs, and for encouraging me to
write on the subject. Among the invaluable resources
provided was a CD-ROM of J. David Rogers’s
research on the failure of the dam. Rogers’s paper in
the book edited by Doyce Nunis, referenced below,
was another valuable source of information and quotations on the history of the project. Finally, I am
grateful to Rogers for his helpful comments on a
draft of this column.
Bibliography
American Society of Civil Engineers, Committee of Board of
Direction. 1929. Essential facts concerning the failure of the
St. Francis Dam. Proceedings of the ASCE 55: 2147–2163.
Bowers, Nathan A. 1928. St. Francis Dam catastrophe—a
great foundation failure. Engineering News-Record,
March 22, pp. 466–472.
Bowers, Nathan A. 1928. St. Francis Dam catastrophe—a
review six weeks after. Engineering News-Record, May
10, pp. 726–733.
Lawrance, Charles H. 1995. The Death of the Dam: A Chapter
in Southern California History. Privately printed.
Nunis, Doyce B., Jr., ed. 1995. The St. Francis Dam Disaster,
Revisited. Los Angeles and Ventura: Historical Society
of Southern California and Ventura County Museum of
History and Art.
Outland, Charles F. 1963. Man-Made Disaster: The Story of St.
Francis Dam, Its Place in Southern California’s Water System,
Its Failure and the Tragedy of March 12 and 13, 1928 in the
Santa Clara River Valley. Glendale, Calif.: Arthur H. Clark.
Rogers, J. David. 2000. Man made disaster at an old landslide dam site. Privately distributed CD-ROM. See also
Rogers in Nunis.
© 2004 Sigma Xi, The Scientific Research Society. Reproduction
with permission only. Contact [email protected].